Classes where average grade % is failing - is this common?

@NeoDymium - that’s what I saying - Test 1 is a bad way to write a test.

Test 1 is a bad test, but it results in a higher average that Test 2. The point is that looking at averages test scores is secondary to other considerations, when evaluating how good a test is.

I’m not sure how you teach people to be better problem solvers other than by showing them how to solve problems and by giving them problems to solve, with some support for doing so while they are working on it and with answers for them to review when they have finished.

No one is saying that humanities are plug and chug. I feel that some people on this thread are just making up accusations to fling around. I have no idea how you would teach creativity in the humanities other than by showing some examples and giving some assignments, and reviewing some examples of outstanding work. I view it as much the same process. It’s the non-STEMy people on this thread who apparently think that STEM is meant to be regurgitation whereas creative thinking and problem solving is only the domain of humanities.

Many of these profs give pretty much the same tests year after year with the same results. It really isn’t unpredictable in that situation to figure out how to improve the teaching and/or the test – it becomes obvious what the students don’t understand. If the profs want that end result…

I go back in my mind to my Calc TA in college – he was distraught when our section (and all others) had this kind of result (low scores across the board on our midterm). He worked like a crazy man to (1) make sure we DID have the teaching we needed to handle those problems on the final, and (2) make sure he knew what the upcoming tests were like so he could prepare us effectively. He added office hours, have us additional problem sets and had review sessions of the result. He was a guy who cared about this issue, and worked to avoid it going forward once it occurred. My Chem TA & prof - ha. Not at all. This is a choice.

I’ll give an example of how to “teach” STEM problem solving. Sometimes (often!) I have to follow the lead of the students, depending where in the curriculum they find something intriguing. It’s my job to teach the starting points, lots of them, and expose them to interesting problems, experiments, etc.

In an Honors science class, I linked to some web-based challenge questions, one of which was accidentally calculus based. But some of the kids tried to solve it anyway, even after I said, “forget problem X because you haven’t had calculus yet”. So since the class seemed interested, I gave a few ideas to try to solve the science problem, on the board:

  1. Write a sentence about what seems to depend on what other variable (e.g. "The rate is proportional to the amount we have left.") [generalizable method - make note of what you do know, in familiar language, and rethink what is being asked]
  2. Try to graph some theoretical data that could look like this [generalizable method - use a tried/true technique to attempt to model the new thing you just learned about]
  3. What kinds of functions in math might behave like your graph? Experiment with the graphing calculator. [generalizable method - be brave enough to guess at what might work; be prepared to fail a few times; use your technology tools as you proceed]
  4. In plain English, what kind of observable behavior might this be describing? [generalizable method - be able to explain your work, and be able to apply your knowledge and communicate about it]

Also, as a HS teacher, I view part of my job as preparing the kids emotionally for the experience of seeing new or scary/difficult STEM problems. You have to work through the feelings of “OMG I have no clue” and “Does everyone else understand this except for me” and “What if the first thing I try, doesn’t work”. The only way to do this is to practice and be coached and reassured - and challenged! And honestly, some kids are not at the level to do this kind of thinking or don’t have the tenacity or eagerness to develop insight. Yes, there can be late bloomers, in all kinds of ways, but sometimes the hard Honors courses are not for everyone.

When someone is going to engineer a new building, or invent a cardiac stent, or design a rocket, I want them to be able to solve a new problem that has not been solved before. I also want them to be able to work in teams, be humble enough to have their work checked by others, be careful enough not to make unit-conversion errors, etc…

In other words, have 80% of the exam devoted to C student problems?

Actually, as noted before, even C students need to be able to apply concepts learned in class to new problems in subsequent classes. Actually, even in elementary school, there was the expectation that one could apply math to problems other than equations in math class. As examples, they gave word problems like “Suppose you have five slices of pizza and then eat two of them. How many do you have left?”.

A well written test will have some C problems, some B problems, and some A problems, for a reasonable definition of C, B, and A grades for the expectations of the course. But the percentage need not be so heavily weighted toward C problems like it is in high school. Note that this is independent of curved grading, since a test with one each of C, B, and A problem (equally weighted) can be graded on a non-curved absolute scale (e.g. 25%, 55%, 85% thresholds to account for minor errors).

If the concern is that students will be discouraged by apparently low percentage scores on tests, even though they may correspond to B (or A) level work, then perhaps instructors need to use the tricks described in replies #97 and #99.

The point is that both of your suggested tests are bad. In the first one, knowledge of certain material is going to be weighted higher than other material. In the second one, specific knowledge on part of the material is going to be weighted more than rudimentary knowledge of everything.

Let’s say the test has four concepts A, B, C, and D. Student 1 knows 80% of each. Student 2 knows 100% of two of them and nothing about the other two. In a standard, somewhat fair system, student A gets a B (he knows most of the material but needs some work) while student B would get a D (knows a lot about part of it but really isn’t qualified to move on because the other two concepts are critical as well).

Test 1:
Student 1 gets pretty high but not perfect marks on the easy ones (say 20 a pop) and pretty low scores on the difficult ones (say 5-10 each) because his/her knowledge isn’t perfect, and it really hurts on the tougher questions and a little bit on the easy ones. Overall, 50-60%.
Student 2 gets a 50%.
Mildly favorable to Student 1 despite being a much stronger student.

Test 2:
Student 1 can’t quite get all of the most difficult parts of the difficult questions, and gets 5-10 points per. Overall ends up with a 30-40%.
Student 2 gets a 50%.
This is even worse.

A better test would test all four of the concepts at both a rudimentary and an advanced level rather than in an uneven matter. Good lecturers do this. Not all professors are good lecturers nor do they want to be.

@NeoDymium I don’t disagree and didn’t mean to imply that Test 2 couldn’t do this. Was only trying to draw a contrast between Test 1 and 2

Frosh-level calculus and physics syllabi seem to be the same as they were decades ago. The frontiers of knowledge in math and physics were far beyond those courses back then, as they are now, so the pushing out of the frontiers of knowledge in those subjects has little or no effect on the frosh-level courses.

Even in the rapidly changing field of computer science, the introductory CS course for CS majors at Berkeley was essentially the same for 25 years, and the current course is a modification of that course.

My professor for genetics once said something along the lines of, “By the time your children take this class, 20% of the material will have been found to be obsolete and way off the mark. But as of now we do not know which 20% that is.”

There were some claims in this thread that women do worse in more competitive (against other students) situations than men do.

But how does that explain the large number of women who go to medical school (where the pre-med process is widely seen as among the most cutthroat-competitive environment in college), or that the biology major (where lots of the cutthroat-competitive pre-meds are) is majority women? Also, how does that explain the situation where women in the computer industry seem to be more likely than men to be immigrants from countries where the application to attend university is sitting for a highly competitive standardized test?

AoPS is a good example of an educational approach that does encourage inductive and deductive reasoning for problem-solving.

Richard Rusczyk shared this quote at Math Prize for Girls:

http://mathprize.atfoundation.org/archive/2009/rusczyk.pdf

This is one of my main motives for homeschooling. My entire educational philosophy is built around developing higher order thinking skills.

I think the point some are making is that this should be shown to them at a point earlier than the exam. In class, in lab, whatever.

Someone did, with an example of French conjugation.

Anyone who is saying that students are not being exposed to problem solving before the exam doesn’t seem to understand that the major activity in most STEM classes, an activity that can easily consume upwards of 10 hours per week, is working through the problem sets. How can you suggest that the students are being hit with a new expectation that they apply the concepts to problems in the exams? What do you think they’ve been doing for many hours every single week? Practicing applying the concepts and equations to solve new problems. What do you think goes on in discussion sections? Going over how to apply the concepts to solve new problems.

I feel the suggestions being made are about the same level as if I were to suggest that college English courses should require no more of students than to provide a plot summary of the books they read in the class. Maybe that’s what goes on earlier in the educational process, but by college a certain amount of creativity and analytical ability is expected.

Statistics for med school applications, acceptance, and graduation here: https://www.aamc.org/data/facts/

Though I don’t entirely accept the gender argument (I think there’s a correlation between being female and the factors that drive people away from STEM careers but that some females aren’t discouraged and that many males are), my guess as to answer to the question of biology enrollment would simply be “mathematics education.” Math education in the US is among the worst in the developed world and it also has a habit of discouraging people who have a hard time to start with. I think young women would have a tendency to be more likely to be discouraged by early failures in mathematics (anecdotally I find this to be the case, that my female friends all agree that they would appreciate math better if they were taught it better). This would discourage them from chemistry, physics, math, and engineering, but not necessarily from biology.

This is not true in all nations (for example definitely not true in the ex-USSR and the issue may even be flipped to some extent) so that would favor immigrant women over US women.

This begs the question…why the average of 50% or whatever on exams if students are so familiar with this?

I actually don’t have a dog in this fight, I have no problem with a test where the average is 50 if that’s passing, I object to curves set to a certain number of failing grades.

But really, why wouldn’t everyone who worked through the P sets and got feedback afterwards, etc be able to successfully do similar ones on an exam?

Because it’s hard to solve problems.

I don’t know of any curve that is pre-set to fail the bottom X students, though I suppose someone may do that. I do know that sometimes students misunderstand or misrepresent how the grading is being performed. Curves are usually set to the mean or median grade. If you look at the distribution of scores there is usually a point below which it seems reasonable to say this is not passing.

Interestingly, on another thread I expressed surprise that only 10 or so students are able to achieve perfect scores on the AP calc tests each year. There are over 400,000 students taking these exams, and perhaps 6 of them can do all the problems. These are more basic problems than the typical college STEM class requires and yet, other posters unanimously thought that it was unreasonable to be surprised that only one or two per hundred thousand students are able to do them all. Yet here, many posters can’t understand why in a class of 300 students, maybe no one can score over 90. Strange.

We get it. Humanities are the same as stem courses. No one course is any easier or more difficult than any other. It’s the new politically correct curriculum. Probably would help if you tell the student body that.

If a young woman (or a young man) gets upset because they got 58% right on a test where the professor has announced that the class average is 40%, they might not be cut out for a STEM major. If they’re so poor at quantitative reasoning, they probably should look to some non-STEM subject.

There’s no advantage to doing them all and studying hard enough to get more than what will secure you a 5, they’re not all simple questions, in a test that long it’s very easy to make at least one mistake. Working harder than necessary for no gain is pointless.

So, it’s “pointless” to study more than what will get you 60% in high school, but you’re outraged that college STEM tests tend to average around 60%?